Quantifying CO2 uptake and biological productivity in the southern hemisphere oceans using atmospheric observations

In this thesis, the use of atmospheric observations for quantifying the uptake of `CO_2` and biological production in the Southern Ocean is explored. For both these processes, ocean observations appear insufficient to accurately quantify their magnitude and variability. It has been proposed that atmospheric observations would provide a better constraint. In the first section, we use an atmospheric inversion model to combine atmospheric and oceanic observations to investigate the southern hemisphere ocean `CO_2` uptake. From sensitivity studies that vary both the initial ocean flux distribution and the atmospheric data used in the inversion, our inversion predicts a total (ocean and land) uptake of \\(1.65\\) to \\(1.90\\) \\(GtC\\) `y^-1`. We assess the consistency between the mean southern hemisphere `CO_2` ocean uptake predicted by an atmospheric inversion model for the 1991-1997 period and the ocean flux estimate based on observed `‚Äöv†vúC0_2`, as in Takahashi et al. (2002). In the Takahashi et al. (2002) paper the ocean flux estimate is referred to as T99, since the data were first released in 1999. In this study, we adopt the same name for this estimate. The inversion cannot match the large `1.8` `GtC` `y^-1` southern extratropical (20¬¨‚àûS - 90¬¨‚àûS) uptake of the T99 ocean flux estimate without producing either unreasonable land fluxes in the southern mid-latitudes, or increasing the mismatches between observed and simulated atmospheric `C0_2` data. The southern extratropical uptake is redistributed between the mid and high latitudes. Our results suggest that the T99 estimate of the Southern Ocean uptake south of 50¬¨‚àûS is too large, and we hypothesize that the discrepancy reflects the inadequate representation of winter-time conditions in the T99 estimate. In the second section, we apply the methods used for estimating biological new production from air-sea oxygen fluxes and atmospheric `O_2`/`N_2` concentrations to a series of model experiments, and evaluate how well simulated new production is retrieved in the southern hemisphere oceans. The air-sea fluxes of oxygen are simulated using a global biogeochemical ocean circulation model. A series of perturbation experiments are designed to span a feasible range of oceanic circulation and biological productivity scenarios. The atmospheric `O_2`/`N_2` fields are simulated by using the `O_2` and `N_2` air-sea fluxes, selected from the global ocean biogeochemical model perturbation experiments, taken as a lower bound in a global atmospheric transport model. Estimation of new production using both the oceanic and atmospheric methods is based on the biological seasonal net outgassing (SNO) of oxygen, which is calculated by integrating the biological oxygen flux over the period where the flux is outgassing to the atmosphere. We demonstrate that the largescale circulation and the disequilbrium-driven fluxes of oxygen produce substantial spatial variability in the g (the proportion of photosynthetically produced oxygen outgassed to the atmosphere), which cause the standard oceanic approach for estimating new production to fail. Fortunately, biogeochemical and physical model perturbations only produce small changes to the spatial structure of the `g` (i.e. ¬¨¬± 0.1 ), since the spatial distribution of the `g` is largely set by the large-scale ocean circulation. The atmosphere acts to integrate the spatial structure in the air-sea fluxes of oxygen. Therefore, it is possible to use a single `g` when applying the atmospheric approach. If we know the hemispheric `g`, then variability in the biological SNO of atmospheric oxygen primarily reflects variability in new production. Using simulated atmospheric `O_2`/`N_2` concentrations, we show that the magnitude of new production can be retrieved if the `g` and the dilution factor (i.e. the proportion of the atmosphere into which the`O_2`/`N_2` signal is well mixed) are known, given that the atmospheric observing network is large enough to capture the spatial variability in the `O_2`/`N_2` signal.